open-llm-leaderboard-old/details_ToastyPigeon__smolphin-test1
收藏数据集概述
数据集来源
该数据集是在对模型 ToastyPigeon/smolphin-test1 进行评估运行期间自动创建的,评估结果展示在 Open LLM Leaderboard 上。
数据集结构
- 配置数量:63个配置,每个配置对应一个评估任务。
- 运行次数:数据集由1次运行创建。每个运行在每个配置中作为一个特定的分片存在,分片名称使用运行的时间戳。
- 训练分片:"train" 分片始终指向最新的结果。
- 结果配置:一个额外的配置 "results" 存储所有运行的聚合结果,用于计算和显示在 Open LLM Leaderboard 上的聚合指标。
数据加载示例
python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_ToastyPigeon__smolphin-test1", "harness_winogrande_5", split="train")
最新结果
以下是 2024-03-25T03:07:39.363881 运行的最新结果: python { "all": { "acc": 0.25243942480267767, "acc_stderr": 0.030541803490214812, "acc_norm": 0.25334593118504556, "acc_norm_stderr": 0.03130353300749519, "mc1": 0.22399020807833536, "mc1_stderr": 0.014594964329474202, "mc2": 0.3581370537867539, "mc2_stderr": 0.01363579277070249 }, "harness|arc:challenge|25": { "acc": 0.29692832764505117, "acc_stderr": 0.013352025976725225, "acc_norm": 0.3225255972696246, "acc_norm_stderr": 0.013659980894277364 }, "harness|hellaswag|10": { "acc": 0.44722166899024096, "acc_stderr": 0.004961904949171386, "acc_norm": 0.5972913762198765, "acc_norm_stderr": 0.004894407257215792 }, "harness|hendrycksTest-abstract_algebra|5": { "acc": 0.26, "acc_stderr": 0.04408440022768078, "acc_norm": 0.26, "acc_norm_stderr": 0.04408440022768078 }, # 其他任务的结果... }
配置详情
-
配置名称:harness_arc_challenge_25
- 数据文件:
- 分片:2024_03_25T03_07_39.363881
- 路径:
**/details_harness|arc:challenge|25_2024-03-25T03-07-39.363881.parquet
- 路径:
- 分片:latest
- 路径:
**/details_harness|arc:challenge|25_2024-03-25T03-07-39.363881.parquet
- 路径:
- 分片:2024_03_25T03_07_39.363881
- 数据文件:
-
配置名称:harness_gsm8k_5
- 数据文件:
- 分片:2024_03_25T03_07_39.363881
- 路径:
**/details_harness|gsm8k|5_2024-03-25T03-07-39.363881.parquet
- 路径:
- 分片:latest
- 路径:
**/details_harness|gsm8k|5_2024-03-25T03-07-39.363881.parquet
- 路径:
- 分片:2024_03_25T03_07_39.363881
- 数据文件:
-
配置名称:harness_hellaswag_10
- 数据文件:
- 分片:2024_03_25T03_07_39.363881
- 路径:
**/details_harness|hellaswag|10_2024-03-25T03-07-39.363881.parquet
- 路径:
- 分片:latest
- 路径:
**/details_harness|hellaswag|10_2024-03-25T03-07-39.363881.parquet
- 路径:
- 分片:2024_03_25T03_07_39.363881
- 数据文件:
-
配置名称:harness_hendrycksTest_5
- 数据文件:
- 分片:2024_03_25T03_07_39.363881
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2024-03-25T03-07-39.363881.parquet**/details_harness|hendrycksTest-anatomy|5_2024-03-25T03-07-39.363881.parquet**/details_harness|hendrycksTest-astronomy|5_2024-03-25T03-07-39.363881.parquet**/details_harness|hendrycksTest-business_ethics|5_2024-03-25T03-07-39.363881.parquet**/details_harness|hendrycksTest-clinical_knowledge|5_2024-03-25T03-07-39.363881.parquet**/details_harness|hendrycksTest-college_biology|5_2024-03-25T03-07-39.363881.parquet**/details_harness|hendrycksTest-college_chemistry|5_2024-03-25T03-07-39.363881.parquet**/details_harness|hendrycksTest-college_computer_science|5_2024-03-25T03-07-39.363881.parquet**/details_harness|hendrycksTest-college_mathematics|5_2024-03-25T03-07-39.363881.parquet**/details_harness|hendrycksTest-college_medicine|5_2024-03-25T03-07-39.363881.parquet**/details_harness|hendrycksTest-college_physics|5_2024-03-25T03-07-39.363881.parquet**/details_harness|hendrycksTest-computer_security|5_2024-03-25T03-07-39.363881.parquet**/details_harness|hendrycksTest-conceptual_physics|5_2024-03-25T03-07-39.363881.parquet**/details_harness|hendrycksTest-econometrics|5_2024-03-25T03-07-39.363881.parquet**/details_harness|hendrycksTest-electrical_engineering|5_2024-03-25T03-07-39.363881.parquet**/details_harness|hendrycksTest-elementary_mathematics|5_2024-03-25T03-07-39.363881.parquet**/details_harness|hendrycksTest-formal_logic|5_2024-03-25T03-07-39.363881.parquet**/details_harness|hendrycksTest-global_facts|5_2024-03-25T03-07-39.363881.parquet
其他路径...
- 路径:
- 分片:2024_03_25T03_07_39.363881
- 数据文件:
}



